I’m using the Xbee to transmit data from two Flexiforce sensors. I’ve checked the hardware multiple times and the signals are not interacting. Checking with a multimeter, the input to ADC0 only changes with sensor 1 change and ADC1 only changes with sensor 2 changes. The voltage to the sensors and the ground signal are stable.
However when I parse the Xbee output signal, where 1 is the indicator value of 126, bytes 12 and 13 are controlled by ADC0 but 13 is also affected by the ADC1 input. Byes 14 and 15 are controlled by ADC1 but 15 is also affected by ADC0 input. The attached figures show what my data looks like in Labview and what my DAQ Xbee settings are.
How large is the ‘error’ (#bits) on the other ADC channel?
When the ‘sensor’ is NOT activated what is the signal on the XBee Analog input pin? Ground? Floating? Vdd?
My guess is that the XBee’s ADC uses its own Vdd (3.3V) for the reference and that Vdd does droop slightly affecting the ADC results.
Series 1 XBees do use Vdd for the ADC Vref. If you watch Vdd on a scope (AC couple & ~10mv/div) I’ll bet you can see a Voltage drop of 10’s to 100’s of mV when the XBee transmits.
Can you connect an external stable Voltage to the XBee’s Vref (pin 14) and try again?
Also study the XBee document (Digi #90000982). Check Table 1-05
The max cross talk error I get is between 2 and 4 bits (max 5 decimal change effect from one sensor and 12 from the other). They are theoretically identical sensors, but I have not calibrated the second one yet, so I can’t be sure.
You are right that the signal is measured relative to Vcc and is high without actuation of the sensor. I tried comparing my Vcc to a reference voltage of 3.3V. I’m still not seeing any change with the multimeter with applied force to the sensors. The oscilloscope is currently being used by others for testing but I will check tomorrow. I agree that this really does feel like a hardware issue.
Sure enough it was hardware. The impedance of the sensors was too high for the Xbee, and resulted in the cross talk issue. Buffers have been added and now the problem is gone.